Azinet Web Site Developer's Reports

The following reports (updated periodically) cover search engine tactics, browser wars, link exchanges, referral data and other tips, hints, and general information for our customers, other web site owners, or anyone interested in web site development. Please send any comments to ted(atsign)

Banner Advertising and Banner Exchanges 1/98

Banner advertising is becoming a major web activity of interest to many site owners. Banners are displayed on a web site and when clicked transport the viewer to a page on another web site. One exposure of a banner is generally referred to as an "impression". If the banner is clicked by the viewer it is a "click-thru". The ratio of click-thrus to impressions is the "click-thru ratio". If the subject of the banner ad is of broad but not universal appeal (Golf for example) and the banner is displayed to a random audience (Infoseek's home page for example) then one could expect a click-thru ratio of about one to three percent. Subjects of more universal appeal (Internet services, computers, etc.) could expect higher ratios. Banner advertising can be purchased in small or large amounts for approximately $.01 per impression. Since this would amount to about $.50 per clickthru it would cost $100 per week to double the viewership of a site that is getting 200 home page visits per week. Random target banner advertising is therefore of most interest for very wide appeal subjects.

Banner exchange programs such as LinkExchange offer to swap banner exposures usually on a 1:2 rate (one exposure of your banner for every two exposures of someone else's banner on your site). If you only display the exchanged banner on your home page your viewership would only increase 1 percent! (.5 exchange ratio X .02 click-thru ratio) Some people use banners on every page and more than one exchange service but this is a lot of trouble and trashes up your site for at most a 10 percent increase in viewership.

An obvious improvement over random banner advertising is "targeted" advertising in which your banner is only shown on sites with a subject similar to your site. People reading a golf site should be much more likely to click on a golf banner than a random audience. LinkExchange recently implemented targeting in their banner exchange service. Unfortunately, click-thru rates actually declined in the targeted mode indicating either an extremely unexpected and unexplained outcome or some kind of problem with their software or approach. We intend to keep following this development.

The most cost effective advertising approach at present is to either swap links or banners with another site with a similar subject or purchase banner advertising space from a high traffic site having the same subject as your site (and reasonable rates!).

To learn more about banner advertising check out Mark Welch's Banner Advertising Information.

E-Mail SPAM Problems and Prevention 12/97

Unsolicited commercial e-mail, also known as SPAM, is a major and growing problem. Many of us are getting ten or more SPAMs a day. Legislative solutions to the problem are in work but may take a long time to happen and may not be effective. There are some methods that can be used by individuals to reduce the SPAM problem and its impact. Approaches are generally to filter the email after you receive it or to adopt methods to reduce the amount of SPAM sent to you.

Many E-mail client programs such as Eudora have "filter" features that can be used to route some messages to "trash" or to route preferred messages to a preferred mailbox. Since SPAM is almost never addressed to you individually you could trash any mail that doesn't list you in the "TO:" header. Other filter techniques can also be used. Unfortunately you still end up downloading the SPAM. Also mail from mailing lists that you want to receive could be unintentionally deleted. You could filter based on known spammer's e-mail return addresses but they typically use fake addresses and change them frequently. Some "responsible" spammers (if that is not an oxymoron) are now using the word "advertisement" or "UCE" (Unsolicited Commercial E-mail) in their messages to allow users to filter out their ads and some legislation would require such tags. Unfortunately, to date, very few spammers use these tags.

Spammers have to obtain your email address from somewhere so the second main approach is to reduce SPAM at its source by reducing the chance that you will appear on a spammer's list.

A major source of email addresses is addresses embedded in web pages. Inexpensive software is available to search the web and extract email addresses. Of course it is essential for most web page sponsors to put their email addresses on their page and the HTML "mailto" tag allows one click customer email response so we are looking at alternatives that would fool the search programs but still allow customers to reach web sponsors. One possibility is to eliminate the mailto tag and disguise the email address as in joe(atsign) or The customer would need to manually edit the address to reach you. Another possibility is to use a form to allow the customer to submit comments or requests. This method could be instead of or in addition to the disguised email address method.

Another source of addresses is return addresses on messages you post to usenet newsgroups. Again, software can automatically extract these addresses for use in sending SPAM.

Some Don'ts: Don't attempt to reduce SPAM by replying when it says "send reply to be removed from our list". Since spammers are by definition sleazy they will probably sell your address to other spammers as a "verified" address. Don't put an entry in AOL's Member Directory - this is guaranteed to get you lots of SPAM. Directories of email addresses such as 411 can also be sources. Don't do business with spammers. Needless to say, doing business with a spammer will put you at the top of their email address list.

Search Engine Data 9/97

The Azinet web server obtains referral data from web browsers accessing the site. That is, the browser sends the URL of the page it was displaying immediatly prior to accessing a page from the server. This page is usually the search engine, directory, or linking site containing a link to your page (the referring site). In the case of some search engines, the URL contains the terms that the user was searching for when the linking page was displayed and that information is also passed to the server and recorded. We use this to develop statistical data regarding reference rates from various search engines and directories as shown below.

The most important engines and directories in order of decreasing total users are:

Note that Excite also provides the data for Netfind and Webcrawler and that AltaVista does the searches for Yahoo (if the user's keywords are not found in the Yahoo directory) and provides the data for Looksmart.

For all Azinet sites combined the referral data is as follows in percent of referrals from each service.

Yahoo 21
Infoseek 12
AltaVista 9
AOL Netfind 8
Lycos 6
Webcrawler 5
Excite 2 1
Hotbot 1
Surfpoint 0.2
LinkExchange 0.03
Purchased Ads 10
Other 24.8
Total 100

The Azinet data agrees fairly well with the service ranking. The "other" category includes referrals from link swaps, manual entry of a site address, remaining search engines and directories, and all other sources.

Search terms provided by some search engines indicate that many site hits were the result of user searches for something completely different from the site content. These people either did an "impulse" click on a site that clearly had nothing to do with what they were looking for or did not read or understand the one sentence blurb that the search engine usually gives users (some search engines can be set to display page titles only). Make sure the title on each page is as descriptive as possible! This is probably the most important single "search engine secret".

Note the major importance of Yahoo. There is some evidence that Yahoo hits are of higher quality (user more likely to actually be interested in site content) than search engine hits. A Yahoo user follows a very structured directory to find your site and should not have the problems mentioned in the previous paragraph. Web site owners should attempt to register their site in all applicable categories and in regional Yahoo directories. Azinet customers should contact Azinet to discuss any questions regarding Yahoo registration.

We are currently mystified as to the poor performance of Excite relative to Webcrawler and Netfind which are said to use the same database. That is, if your page is registered with Excite it should show up on Netfind and Webcrawler.

Browser Wars 8/97

Azinet collects statistics on web browsers used to access our site. As of August 1997 we see the following statistics:

The "Mozilla" listings include Netscape AND Microsoft Internet Explorer so this shows that Netscape has 59 percent and Microsoft has about 38 percent. Web TV is still insignificant. Despite the fact that the new Netscape Communicator (Mozilla 4) has been out for a while it doesn't seem to be picking up much. Microsoft seems to be picking up steam. Macintosh seems to be fading fast.

AZINET Browser Agent Analysis 08-30-1997 Browser Summary

        0.00         Mozilla/0
        2.62         Mozilla/1
       50.50         Mozilla/2
       37.67         Mozilla/3
        6.36         Mozilla/4
        0.00         MSIE 1
        3.89         MSIE 2
       33.48         MSIE 3
        1.12         MSIE 4
       14.07         AOL
        0.55         WebTV
User Operating System Summary
       61.97         Win95
       27.73         Windows 3.1
        4.54         Macintosh
        1.33         X11
        0.41         SunOS
        0.17         SCO
        0.02         Linux
        4.04         WinNT
        0.32         HP-UX
        0.15         IRIX
        0.07         Amiga

Search Engine Secrets 8/97

Have you been getting a lot of e-mail offering to sell search engine secrets? We sure have. So here is some free information.

Search engine crawlers are generally supposed to eventually find any page on the web that is linked from any other page that they already have. That is, if you give one your home page address they should register all your pages. Initially, they actually did this until it became obvious that a lot of complete trash was being indexed and they were filling up their disk space. Now, apparently, most search engines only index pages that are actually submitted despite claims to the contrary. This saves them disk space and should result in a better search experience since pages that nobody wants registered will eventually disappear. The engines are supposed to periodically check the pages in their index to remove pages that no longer exist and update information on pages that have changed. It is not clear that they are actually doing this very well either based on the number of dead links we are seeing in search engine results pages.

The upshot is that web site owners need to check their pages periodically to see if they are still registered and reregister them if they are missing. If one of your registered pages changes in a way that would affect searching it needs to be resubmitted.

You can find out if your pages are indexed by doing a search on your company name or some other unique word or phrase on your pages.

Should you register every page on your site? There is some benefit to registering every page since somebody might search for a word or phrase that happens to be on a particular subpage and not on the home page. Usually there are two or three most important pages in a small site and it may not be worth it to register and maintain registration for the other pages unless you feel they will get significant additional hits. If you do register a page make sure of the following:

The page should have a title (HTML title tag) as descriptive as possible. The first few words on the page should be as descriptive as possible to a search engine user while still looking good to a reader of the whole page. This is frequently quite difficult to do. I have seen 300 page web sites where search engine users see garbage when looking at the HOME PAGE! Check what your pages look like to a search engine user when you check for indexing.

Web Site Developer's Reports provided by AZINET Business Internet Solutions ( 1997